FastGauss: improve javadoc

Author: maarzt

src/main/java/net/imglib2/algorithm/convolution/fast_gauss/FastGauss.java

@@ -23,12 +23,14 @@
  * Faster alternative to {@link net.imglib2.algorithm.gauss3.Gauss3}.
  * It's especially faster if sigma is larger than 3.
  * <p>
- * It is normally expensive to calculate the exact result of the gauss convolution.
+ * It's expensive to calculate the exact result of the gaussian blur.
  * That's why approximations are almost always used.
  * {@link net.imglib2.algorithm.gauss3.Gauss3} does an approximation
- * by cutting of the gauss kernel at 3*sigma + 1.
- * {@link FastGauss} uses a different approach known as Fast Gauss
- * Transformation. The runtime is thereby independent of sigma.
+ * by cutting off the gauss kernel at 3*sigma + 1.
+ * {@link FastGauss} uses a different approach known as fast Gauss
+ * transformation. It's runtime is independent of sigma.
+ * <p>
+ * See {@link FastGaussCalculator} for more details.
  *
  * @author Vladimir Ulman
  * @author Matthias Arzt

src/main/java/net/imglib2/algorithm/convolution/fast_gauss/FastGaussCalculator.java

@@ -1,7 +1,14 @@
 package net.imglib2.algorithm.convolution.fast_gauss;
 
 /**
- * Fast algorithm to calculate an good approximation of the gauss convolution.
+ * This class implements the fast Gauss transform to calculate a gaussian blur.
+ * The approach is very different from an algorithm using a truncated convolution kernel.
+ * Especially the runtime is independent of sigma.
+ * <p>
+ * The implemented algorithm is described in detail in:
+ * Charalampidis, Dimitrios. "Recursive implementation of the Gaussian filter using
+ * truncated cosine functions." IEEE Transactions on Signal Processing 64.14
+ * (2016): 3554-3565.
  *
  * @author Vladimir Ulman
  * @author Matthias Arzt

src/main/java/net/imglib2/algorithm/convolution/fast_gauss/FastGaussConvolverRealType.java

@@ -9,7 +9,7 @@
 
 /**
  * Implementation of {@link LineConvolverFactory} that uses {@link FastGaussCalculator}
- * to calculate a fast gauss convolution.
+ * to calculate a fast Gauss transform.
  *
  * @author Vladimir Ulman
  * @author Matthias Arzt